How Many Miles Is Too Many for a Used Car?

Ask any group of people how many miles is too many for a used car and you will get a number back within about two seconds. Usually it is 100,000. Sometimes 150,000, from someone who wants you to know they are not superstitious. Occasionally 60,000, from someone who has been burned.

What none of those answers has is a source. The 100,000-mile line is a genuine piece of folk knowledge — it was a reasonable rule when carburettors, points ignition and iron-bore engines meant most cars were genuinely finished somewhere in six figures. It survived the engineering that made it obsolete, the way rules of thumb do, and it now sits in the middle of the used market as an invisible price cliff that has very little to do with the cars on either side of it.

This page tries to do something different from repeating a number. It goes to the largest public record of things going wrong with cars in the United States — the complaint database NHTSA maintains — and asks what the odometer actually said when owners reported a failure. The answer is not a threshold. It is more useful than a threshold, and it is also more uncomfortable, because it says the number on the dashboard is a much weaker signal than the way we shop implies.

The threshold nobody can source

Start with what the 100,000-mile rule would have to be true for. It would need failures to be rare below the line and common above it — some mechanism that holds a car together for a defined period and then stops. Cars do not work that way and never really did. They are assemblies of several thousand parts with wildly different service lives, some of which are consumed by distance, some by time, some by heat cycles, and some by whether anybody bothered to change a fluid.

A water pump has a life measured in distance. A rubber coolant hose has one measured mostly in years and heat cycles. A battery does not care about your odometer at all. A timing chain tensioner may last the life of the car or may fail early because of an oil specification the previous owner ignored. Aggregate all of that and you do not get a cliff. You get a spread — and the practical question for a buyer is not where the cliff is but where in that spread this particular car sits, which is a question about its history rather than its mileage.

The market, meanwhile, prices as though the cliff were real. That is worth knowing in its own right, because a rule that is widely believed and weakly supported is exactly where value hides. A well-documented car that has just crossed a psychologically important number is cheaper than the same car a few thousand miles earlier, and mechanically it is the same car.

What 1,324 federal complaint records actually say

NHTSA takes safety complaints from vehicle owners and publishes them. Most of each record is free text, but complaints taken by the agency’s intake staff carry one field written to a fixed template: the odometer reading at the time of the failure, recorded in a standard sentence. That sentence is machine-readable in a way the rest of the narrative is not, so it can be extracted without anybody having to interpret what an owner meant.

We took the archived complaint pull that already backs the vehicle data on this site — 9,096 complaints across a set of 2015, 2016 and 2018 vehicles — and parsed that reading wherever it appeared. It appeared in 14.6 per cent of them, giving 1,324 odometer readings attached to a reported safety failure. Readings that were implausible as odometers were discarded rather than cleaned up.

Here is the distribution.

The middle of that distribution sits at 60,000 miles. A quarter of the readings are below 30,000. The 75th percentile is 93,685, meaning three-quarters of everything reported had happened by the point most buyers treat as the danger line.

Half of it happens before 60,000 miles

The single most useful figure on this page is that 659 of the 1,324 readings — 49.8 per cent — are below 60,000 miles. Nearly a third, 29.4 per cent, are below 36,000: that is, the failure was reported while the car was still inside the mileage limit of the basic warranty it was sold with.

That is not a claim that low-mileage cars are worse. It is a claim about what mileage does and does not tell you. A car with 40,000 miles on it has not proved anything about its reliability; it has simply not been asked many questions yet. When the federal record shows that half of the reported safety failures on a set of ordinary cars arrived before 60,000 miles, the implication for a buyer is direct — a low odometer reading is not evidence of a healthy car, and treating it as evidence is how people pay a premium for a problem.

The inverse matters too. A car at 120,000 miles with an unbroken service history has demonstrated something. It has been through the window in which half of these failures showed up, and it is still here. That is real information, and the market gives you a discount for it.

What this data cannot tell you

There is a limit on this dataset that most pages built from complaint counts quietly ignore, so it goes here rather than in a footnote.

The vehicles in the pull are overwhelmingly from the 2018 model year, and the complaints were retrieved in 2026. That is about eight years on the road. Very few of these cars have yet been driven to 200,000 miles, so almost no failures can be recorded there — not because cars stop failing at that point, but because the cars have not arrived at that point. The data is right-censored, and the thin upper end of the chart measures the age of the fleet rather than the durability of the vehicles. In the chart above, those bands are drawn in grey for exactly that reason.

The highest reading in the set is 255,000 miles, and the 90th percentile is 125,326. Both are reported here for completeness. Neither should be read as a life expectancy.

Different parts fail at different points

Comparisons within the dataset escape that limitation entirely. Every component category is drawn from the same cars over the same window, so they are all censored identically — which means the order in which they appear is a fair comparison even though the absolute figures are held down.

That ordering turns out to be stable and quite informative.

Complaints about vehicle speed control cluster earliest, with a median reading of 27,000 miles. Service brake complaints and forward collision avoidance systems follow. At the other end, complaints about engine and engine cooling have a median of 98,000 miles and petrol fuel systems 89,000.

The shape of that makes sense once you see it. The early cluster is dominated by electronics, software and driver-assistance hardware — things that are either right or wrong from the factory and that reveal themselves quickly. The late cluster is dominated by the things that genuinely wear: cooling systems that have been through enough heat cycles, fuel components that have seen enough fuel. Electrical system and power train complaints sit in the middle at 49,000.

For a buyer this is more actionable than any single threshold, because it tells you what to have inspected on the car in front of you. On a low-mileage car, the risk sits in electronics and driver assistance, and the correct check is whether every system actually works and whether any related recall is outstanding. On a high-mileage car, the risk sits in cooling and fuel delivery, and the correct check is a cooling system pressure test and a look at what has already been replaced.

The same model year, very different mileages

Thirteen models in the pull had enough readings to be worth reporting individually. Because nearly all of them are the same model year, they have been exposed to the road for the same length of time, which makes them directly comparable to each other.

Two things stand out. The first is that not one of the thirteen has a median above 100,000 miles — the highest is 95,000. Whatever the 100,000-mile rule is describing, it is not the point at which these cars started giving their owners trouble worth reporting to a federal safety regulator.

The second is the spread between models. Three-quarters of the Nissan Altima readings are below 60,000 miles; fewer than a third of the Kia Optima readings are. Both are 2018 midsize sedans that have been on the road for the same length of time. Whatever separates them, it is not distance travelled, and no mileage rule applied at the point of purchase would have distinguished between them.

A caution on reading that table too hard: complaint volume follows how many of a model were sold and how motivated owners are to file, so the counts in the second column are not a reliability ranking. It is the medians and the shares that carry the information, because those are internal to each model.

Miles are one clock. There are others

A used car has at least three clocks running, and only one of them is on the dashboard.

Distance consumes the things that rotate, slide and rub: tyres, brakes, bearings, clutches, suspension bushes, the drive belt.

Time consumes everything made of rubber and everything holding a charge, whether the car moves or not. Coolant hoses, engine and transmission mounts, suspension bushes again, the battery, the tyres — a tyre with plenty of tread and a date code from nine years ago is a tyre at the end of its life. A car that has covered 20,000 miles in eight years is not a low-mileage car in any useful sense; it is an eight-year-old car with a low reading, and it will need a batch of age-related work that a higher-mileage car of the same age has usually already had done.

Use is the clock nobody records. A hundred thousand motorway miles at steady temperature is gentler than thirty thousand miles of short cold trips that never bring the oil up to temperature, and both are gentler than a truck that has spent its life towing at maximum rated weight. Two identical readings can mean entirely different amounts of wear, which is a large part of why mileage alone predicts so poorly.

Worth noting that heavy-duty pickups are heavily represented in the dataset above, and a work truck is precisely the case where the odometer under-describes the wear. That is not a flaw in the data so much as a reminder of what the data is measuring.

The questions that replace “how many miles”

If mileage will not answer the question, something has to. These will, and unlike a threshold they are checkable before you spend money.

• Does a service history exist, and is it continuous? This is the strongest single predictor available to a buyer, and it is the one most often skipped because it is boring. Stamps, invoices, a workshop that remembers the car. A gap of two or three years in the middle of the record is worth more attention than 40,000 extra miles.
• Has the scheduled work that matters actually been done? Not all services are equal. Transmission fluid, coolant, brake fluid and, where fitted, a timing belt are the items whose neglect turns into a four-figure bill. Ask which have been done and when, and ask to see it rather than hear it.
• Is there an unrepaired safety recall? Free to check, free to fix, and outstanding far more often than people expect on a car that has changed hands a few times.
• What does the title record say? A salvage, flood or lemon-buyback brand changes what the car is worth far more than any mileage figure, and a brand issued in one state does not always follow the car into the next.
• Does the odometer agree with the record? Discussed below, because it is the one question that makes all the others conditional.
• What does an independent inspection find? A pre-purchase inspection costs a small fraction of any of the repairs it might find, and the component data above tells you what to ask the inspector to concentrate on.

Notice that none of those questions has a mileage in it. That is the point. Every one of them distinguishes between two cars showing the same number on the dashboard, which is exactly what a mileage rule cannot do.

Checking the history before the number matters

Most of the questions above resolve against records rather than against the car. Recalls come from NHTSA’s free VIN lookup. Title brands and prior odometer readings come from the title history, which is assembled from state motor vehicle records. Service history comes from the seller, and its absence is itself an answer.

Running the history first also changes the order of the conversation productively. A car that turns out to carry a branded title or an unrepaired recall is a car you have learned something about before you have spent anything on an inspection, and a car that comes back clean is one worth paying an independent mechanic to look at. Doing it the other way round means paying for inspections on cars that a free federal lookup would have removed from the list.

If you want the full sequence rather than a summary, our guide to checking a used car’s history sets out each record source and what it does and does not cover. For a single car you are seriously considering, a VIN history report pulls the title, brand and reported-mileage records together in one place.

When the mileage itself is the thing that is wrong

Every argument on this page assumes the odometer is telling the truth. Often enough it is not, and the incentive is obvious: mileage moves price, and on a digital cluster the reading is a stored value in a module rather than a physical set of wheels that has to be dismantled.

Federal law requires the seller to disclose the mileage in writing at the point of transfer, and knowingly disclosing a false figure carries statutory damages. That disclosure requirement is what makes the fraud detectable, because each transfer leaves a dated reading in the record. A car whose reported readings climb steadily for years and then step backwards has been tampered with, and the record shows it even when the dashboard does not.

This is the one check that has to come first, because a rolled-back odometer does not merely misprice the car — it invalidates the entire mileage conversation you were about to have about it. Our guide to checking mileage by VIN covers where those readings come from and how to read a sequence that does not add up.

So what is the answer

The answer is that “how many miles is too many” is the wrong shape of question, and the reason it persists is that it is the only question about a used car that can be answered from the listing photograph.

The federal complaint record does not describe a threshold. It describes a long spread in which half of what goes wrong has already gone wrong by 60,000 miles, in which nearly a third of it arrives inside the original warranty, and in which no model examined had a median above 100,000. Set against that, a car with 130,000 documented, serviced, single-owner miles is a better proposition than a car with 45,000 miles, no paperwork and a title you have not looked at.

Buy the history. The mileage is one line in it.

Common questions

Is 100,000 miles too much for a used car?

Nothing in the federal complaint record marks 100,000 as a boundary. In the sample examined here, 78.1 per cent of reported safety failures had already occurred below it, and every one of the 13 models with a usable sample had a median reported-failure reading under it. A well-documented car above 100,000 miles is frequently a better buy than an undocumented one below it, and it will be cheaper because the rule is widely believed.

How many miles is too many for a used car?

There is no such figure, and any specific number offered as one is folk knowledge rather than evidence. What matters is the relationship between the mileage, the age, the type of use and the documented history. A car that has covered a lot of gentle motorway distance with every service recorded is in better condition than a low-mileage car that has sat, or that has been used exclusively for short cold journeys.

Is a car with 150,000 miles worth buying?

It can be, and the question to ask is what has already been replaced. A car at that point should have a documented history of the major wear items — cooling system, brakes, suspension components, fluids, and a timing belt if it uses one. If those have been done and there is paperwork, much of the expensive work is behind it. If nobody can tell you, price the car as though all of it is still ahead.

Is it better to buy a high-mileage car or an old low-mileage one?

Distance and time damage different things. A low-mileage older car still needs the age-related work — rubber, hoses, mounts, battery, tyres regardless of tread — and often has not had it, because the low reading persuaded its owner that nothing was due. A higher-mileage car of the same age has usually already had that work done. Neither is automatically better, but the low reading should not be treated as a discount on the age.

Do high-mileage cars always cost more to run?

Not always, and the assumption skips over what the complaint data shows about early failures. Nearly a third of the reported failures in this sample arrived before 36,000 miles. The variable that actually predicts running costs is maintenance history rather than distance: a neglected car is expensive at any reading, and a maintained one is predictable at most of them.

What mileage should I look for when buying a used car?

Rather than targeting a reading, work out whether the reading is consistent with the car’s age and use, then spend your effort on the records. If the car has covered far less distance than its age suggests, ask why and expect age-related work. If it has covered far more, ask what kind of miles they were and what has been replaced. The reading is a prompt for questions, not an answer to one.

Can I trust the mileage shown on the dashboard?

Not on its own. Digital odometers store the reading in a module, which makes altering it a software task rather than a mechanical one. Federal law requires a written mileage disclosure at each transfer, and those dated readings accumulate in the title record, so a reading that steps backwards over time is visible in the history even when the dashboard looks plausible. Check the sequence before you rely on the number for anything else.